Ceramic Capacitor Polarity

[jaylar41]

Hello,

I've got a little problem with a multi-layer ceramic capacitor that I'm trying to use. I'm using some as filter capacitors on a buck regulator circuit. The noise is filtered just fine, but it seems that too much current is conducting through them during their charge cycle (1st half of the arch on half sine wave). The voltage is fully rectified.

If I change the capacitor to an electrolytic, the current moving through it is attenuated significantly. I believe this is due to the larger ESR on the electrolytic, but when I physically add series resistance to the MLCC, the problem is only partially alleviated. The electrolytic is also polarized so that current can only travel one way. The ceramic capacitor, because it's non-polarized, is discharging straight to ground when it is supposed to be charging.

Why don't I use the electrolytic? Because the application doesn't not call for components that will freeze (or become too hot and explode for that matter). I'm trying to build everything without the electrolytic, period. There doesn't need to be a real nice voltage on the line, hence I don't need bulk capacitors. The line does need to be filtered though, and that's where the MLCCs come in.

My question is this: Is there a way to make my MLCC act more like an electrolytic besides adding a little ESR to it? Is there a way to make my MLCC polarized (add simple components in series, parallel etc...), such that current only travels one way through it?

This is my first time here, so I really don't know what to expect for responses. Any help would be greatly appreciated though. I will be happy to explain more if there's a need to. Thank you very much in advance.

Jason

 

[Skogsgurra]

That is an interesting question, jaylar.

A multilayer ceramic capacitor *should* be non-polarized. And it *should* also have a quite low ESR - one doesn't normally think ESR and MLCC.

How much is "too much" current? And if it is only during the first half-wave, it is probably dependent on when you switch on. Switching on on top of the sine will produce lots of transient current while swithing on closer to zero wil produce less current.

Have you made sure that you switch on at the same angle when you compare measurements? I think that different switch-on angles is the main reason for the different current peaks.

Best way to find out is to use persistence mode on a digital scope and switch on several (tens and hundreds) times. That will give you an idea of the spread.

Gunnar Englund

http://www.gke.org

 

[OperaHouse]

Have you used any software like MURATA Chip Capacitor Data Library to calculate expected results of your application? You may have to use multiple caps in order to handle the current. Assume you are having a temperature problem. Are both capacitors the same value? An answer requires more details.

 

[Comcokid]

Also, since you are replacing electrolytics, you are probably using large-value ceramics. What kind of ceramic? X7R? Z5U? Y5V? Something else? What value capacitor?

With a buck regulator, and possible high-switch frequencies, the capacitor charcteristics require looking at more information than just ESR. What (approximate) switch frequency?

 

[jaylar41]

Thanks for the tips skogs.

"Too much current" is as little as 44mA. It doesn't seem like much, but when zero is wanted, then.... The resistance is added to the MLCC as a snubber.

As far as when the switching takes place: I am switching throughout the rectified signal. It's a PWM scheme that is on for 20% duty cycle, with 300us pulses (off for 1.5ms). With or without the pulsing though, the current draw out of the 18VAC transformer shows this 44mA. When the pulsing is added again, the 44mA is there, but also with a pulsed current (sort of offset with the 44mA).

The reason this is so important to me is that I am trying to get the output current of the transformer to closely resemble the voltage waveform. By the sounds of it, you are probably familiar with PFC, right? If I can average these pulsed currents, and make it resemble a sine wave that is in phase with the voltage, then my job is almost complete.

The advice you give on the switching times may not be true. During low times of voltage, there is a larger dv/dt. Thus, the old capacitor equation would hold true: I=C dv/dt, and more current would pass, correct? When we are at the top of the sine, it's almost flat (if you zoom in far enough) and there is less dv/dt.

I'm not really sure what you mean about the switching angles. Sorry for the confusion, but are you refering to phase shift? Also, I'll have to read my manual about the persistence mode on my scope and see if that will help. Thanks for your reply, I appreciate it. It may be easier to show you a picture. Is there a way to attach pics here? I don't see where I can.

Jason

 

[jaylar41]

I have not used any software to simulate. That's something I'll have to do eventually I'm sure.

The setup is this: I'm using 2 2.2uF ceramic caps with an X7R dielectric. As I said before, these are strictly for filtering purposes. The frequency I'm filtering is approximately 300kHz, and the MLCCs do a great job. It's just that since they are not polarized, they are discharging to ground when they shouldn't be.

Two main reasons for not using the electrolytic: Temperature stability and PCB space. Ceramics are much better at both, and I'd like to keep them. Simply, I need to find out if there is a way to, in effect, polarize a non-polarized capacitor. I thought I read this somewhere on the web, but can't find it now. Thanks for the input.

Jason

 

[Skogsgurra]

This is what I mean: When you switch on, you have zero V on the capacitor and a certain voltage after the rectifier. If you switch on at angle 0, your line voltage is zero and your transient current will be low. If you switch on at angle 90, you have peak line voltage and the switching transient will be at its maximum.

I did understand that you mean capacitor current during the first half sinewave. I may have been mistaken - can you describe what the current looks like?

Yes. I am familiar with active power factor correction. And, yes. I agree about the dv/dt. But I do not think that that is the problem.

Gunnar Englund

http://www.gke.org

 

[jaylar41]

The current is measured coming out of the transformer. When there shouldn't be any return current, I see an approximate step function to about 44mA, and after that, it decreases steadily to zero. The zero current is reached at the peak of the sine wave. Then, on the falling edge of the sine, I see regular pulsing as it should be, without any of that 44mA offset current.

It would be really nice if I could attach a snapshot of my scope. I must not be able to do that here?

 

[Skogsgurra]

Sorry. I got you wrong. Forget what I said about mains.

Look at faq238-1161 for posting pictures and diagrams.

Gunnar Englund

http://www.gke.org

 

[jaylar41]

Here's a picture that will help

What's going on is there is usually 10 pulses of current. In this one though, I've skipped the first 3 illustrating the current being drawn even without me pulsing my circuitry.

Can you see how this would affect the result I am seeking. An average of the pulses should be similar to a sine wave, thus making the power factor closer to unity.

Of course this is a hunch of mine. Do you think this is possible? What I mean is this: Will these pulses average out to a sine wave? Maybe I am wrong and wasting my time, but it makes sense to me.

Thank you.

 

[itsmoked]

Schematic?

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[BrianG]

In following this thread I am surprised that no-one has commented on this before:

Quote from first post: "The electrolytic is also polarized so that current can only travel one way. The ceramic capacitor, because it's non-polarized, is discharging straight to ground when it is supposed to be charging."

Where did this strange idea come form? Current can flow "in" or "out" of any capacitor, i.e. charge and discharge. Electrolytic types are only marked as polarised to ensure that you don't use them on a.c., or so that on d.c. reversal of voltage does not break down the oxide dielectric insulation layer. They DO NOT have a diode property implied in the first post.

 

[itsmoked]

Oh we noticed... That and several other things.. Like, " such that current only travels one way through it?".

That's why we need a schematic...

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[jaylar41]

Hey guys,

here's a basic schematic of what I'm working with.

img[

http://i8.tinypic.com/24xglfr.jpg

]

I'm not sure if this is too vague or not. One piece of advice I've read about is that for regulator circuits like this, the designer shouldn't use two ceramics (or non-polarized caps). There seems to be many schematics with one of each, but there is not many explanations as to why.

By the way, thanks for correcting my mistake about the polarized caps. That was one reason I was reluctant to post things on forums. I'm a very young engineer who just graduated college, and I've got a lot to learn. I've only just started designing, so compared to the pros at these sites, you may say I'm a bit wet behind the ears. That's really something I shouldn't worry about though because I think there's a lot to learn at these forums. There's probably other things I've been mistaken on, so fire away...

Thanks again.

 

[Skogsgurra]

I see no problem with the current waveshape that you get from that diagram. The controller does what it should - but you can surely make it do it better once you understand loop dynamics.

A couple of questions: 1) What is the extra diode doing there? 2) Why are you trying to avoid electrolytics? Is it a fact or ruour-driven decision? Or do you just dislike them?

Using one ceramic parallel to an electrolytic is good engineering practice. It is done to create an HF path so that the high ESR in the electrolytic doesn't cause problems. Rather the opposite to what you were saying in the OP.

Also, this has very little resemblance to a PFC. In such a circuit, you shall take the current reference from the rectified but non-smoothed mains while the power section shall be fed from the capacitors. Is that, perhaps, what the extra diode was supposed for? In that case, connect current reference just after the bridge.

Gunnar Englund

http://www.gke.org

 

[Skogsgurra]

ruour-driven = rumour-driven

Gunnar Englund

http://www.gke.org

 

[jaylar41]

The extra diode probably should not have been included with this schematic. I'm using it to separate the rectified signal from the capacitor smoothed signal. I need the rectified signal for some other circuitry on the board.

I'm avoiding electrolytics because the application is temperature sensitive. As I understand, electrolytics do not perform very well in extreme cold or hot. I've read much about them blowing up, thus causing reliability issues. They also dry out over time which causes other problems. I've been looking into using tantalum, but the price may be too high. I may not have a choice by the end of all of this though.

The PFC method that I am trying to employ in my circuit is not a traditional method where I would sense the current out of rectifier. Instead, current through the load is sensed and controlled in such a way that the current drawn out of the transormer should resemble a sine wave (or pulses that average to a sine). I'm still not positive that this approach is a great one. Will this pulsing effectively improve the PF?

I apologize for not including the whole circuit. But I'm trying not to give away too much information that I've worked quite hard on for quite a while.

Jason

 

[Skogsgurra]

Sorry to be sarcastic. But you should not overestimate the value of chips falling from your bench.

The PWM method is what is used for PFC. It is also used in the active front end in large drives to make current drawn sine formed. It has been done for many years and there is nothing peculiar about that. The current drawn is usually pure (1 or 2 percent THD) sine.

The tales told about electrolytic capacitors may or may not be true. We have electrolytics in cars up here in northern Sweden - sometimes down to -40 C - and no problems. Do get some up to date information and do not believe in old myths. Modern electrolytics are quite good, really. There were some back in the sixties that caused problems in high temperature environments but as long as you use them within specs there is no problem with the modern ones.



Gunnar Englund

http://www.gke.org

 

[geekEE]

How is this circuit being prototyped? Is there a PCB? Perfboard? Protoboard? Parts tied together with clip leads? Schematics don't always show the whole story.

 

[jaylar41]

The prototype is on PCB for the most part. At the moment, I've got some wires coming off of the board and connecting to a microcontroller on a breadboard. I'm doing this so that I can re-program the uC often. The connection to the power supply is indeed made with alligator clips right now.